Evaluation of statistical differential analysis methods for identification of senescent cells using single-cell...

Area of Science:

• Genomics
• Bioinformatics
• Computational Biology

Background:

• Differential gene expression (DGE) analysis is vital for identifying senescent cells in single-cell RNA sequencing (scRNA-seq) data.
• The performance of DGE methods, especially within the Seurat package, requires thorough evaluation.

Purpose of the Study:

• To systematically assess and compare the performance of 10 DGE methods available in Seurat.
• To identify the most effective DGE method for analyzing scRNA-seq data, particularly for senescent cell identification.

Main Methods:

• Evaluated 10 DGE methods (Wilcox, Wilcox-limma, bimod, roc, t, negbinom, Poisson, LR, MAST, DESeq2) using simulated and real scRNA-seq datasets.
• Assessed method performance across varying sample sizes, sparsity levels, and proportions of true differential expression.
• Utilized metrics including false discovery rate (FDR), sensitivity, specificity, accuracy, AUC, and AUPRC.

Main Results:

• DESeq2 consistently outperformed other methods across all tested conditions.
• DESeq2 achieved the highest Area Under the Curve (AUC) and Area Under the Precision-Recall Curve (AUPRC).
• Performance varied based on sample size, sparsity, and the proportion of truly differentially expressed genes.

Conclusions:

• DESeq2 is recommended as the preferred method for DGE analysis in scRNA-seq data.
• The findings provide valuable guidance for researchers selecting DGE methods for scRNA-seq analysis.